Queries on JSONB

[ghost]

this works:

cqlsh> SELECT * FROM store.books where details->'yo'->'shi'->0->>'type' = 'a';

 id | details
----+----------------------------------------------------------------------
 44 | {"yo":{"shi":[{"type":"a","value":"aa"},{"type":"b","value":"bb"}]}}

it would be great if I could target type:b without knowing its index which is 1:

cqlsh> SELECT * FROM store.books where details->'yo'->'shi'->*->>'type' = 'b';

 id | details
----+----------------------------------------------------------------------
 44 | {"yo":{"shi":[{"type":"a","value":"aa"},{"type":"b","value":"bb"}]}}

it could be even better if I could target directly anything embedded without knowing its path:

cqlsh> SELECT * FROM store.books where details->*->>'value' = 'bb';

 id | details
----+----------------------------------------------------------------------
 44 | {"yo":{"shi":[{"type":"a","value":"aa"},{"type":"b","value":"bb"}]}}

[kmuthukk]

hi @pouledodue

Yes, seems reasonable that at some point we extend the search/match capability within the JSON document to support some wildcard expressions.

One note: currently the queries as specified will pretty much result in a full-table scan because currently we don't yet allow creating secondary indexes on JSON attributes.

JSON support and "secondary indexes for primitive type columns" are both features currently in Beta that we expect to GA in the 1.1 release of YugaByte.

In the next phase (i.e. post 1.1), we hope to also allow the ability to create secondary indexes on JSON attributes.

regards,
Kannan